Translation initiation with exotic amino acids using EF-P-responsive artificial initiator tRNA

Abstract Translation initiation using noncanonical initiator substrates with poor peptidyl donor activities, such as N-acetyl-l-proline (AcPro), induces the N-terminal drop-off-reinitiation event. Thereby, the initiator tRNA drops-off from the ribosome and the translation reinitiates from the second amino acid to yield a truncated peptide lacking the N-terminal initiator substrate. In order to suppress this event for the synthesis of full-length peptides, here we have devised a chimeric initiator tRNA, referred to as tRNAiniP, whose D-arm comprises a recognition motif for EF-P, an elongation factor that accelerates peptide bond formation. We have shown that the use of tRNAiniP and EF-P enhances the incorporation of not only AcPro but also d-amino, β-amino and γ-amino acids at the N-terminus. By optimizing the translation conditions, e.g. concentrations of translation factors, codon sequence and Shine-Dalgarno sequence, we could achieve complete suppression of the N-terminal drop-off-reinitiation for the exotic amino acids and enhance the expression level of full-length peptide up to 1000-fold compared with the use of the ordinary translation conditions.

In canonical transla tion initia tion in prokaryotes, P-site N -f ormylmethion yl-tRN A ini (fMet-tRN A ini ) is recognized by one of the initiation factors, IF3, to survey the stability of codon-anticodon interaction and thereby relocated to the acti v e position (24)(25)(26). If IF3 fails in the relocation, fMet-tRNA ini e v entually dr ops-off fr om ribosome and translation reinitiates from the second aminoacyl-tRNA at the A site to gi v e a truncated peptide lacking the N-terminal fMet, which is called a reinitiated peptide (RiP) ( 27 ). This e v ent is r eferr ed to as the N-terminal drop-of f-reinitia tion. Incorporation of noncanonical substrates with poor peptidyl donor activity, such as N -acetyl-L -proline (AcPro), causes mor e fr equent drop-off-r einitiation ( 27 , 28 ). Ther efor e, it is far more difficult to introduce such exotic amino acids at the N-terminus.
The drop-of f-reinitia tion e v ent also occurs in translation elongation when introducing consecuti v e Pro residues into nascent peptides ( 29 ). The low peptidyl donor / acceptor abilities of Pro cause ribosome stalling, where EF-G triggers mistranslocation of P-site peptidyl-Pro-tRNA and Asite Pro-tRNA toward E and P site, respecti v ely. Then, transla tion reinitia tes from the P-site Pro-tRNA to gi v e a truncated peptide lacking the N-terminal region. Howe v er, it is known that the specific translation factor, EF-P, accelerates peptide bond formation between the consecuti v e Pro r esidues, ther eby alleviating the ribosomal stalling (Figure 1 A) ( 30 , 31 ). We have reported that EF-P can also prevent the drop-of f-reinitia tion e v ent in elongation induced by consecuti v e Pro ( 29 ). Since EF-P functions not only in elongation but also in initiation (32)(33)(34) (Figure 1 B), here we hypothesized that the N-terminal drop-of f-reinitia tion can also be alleviated by EF-P. If this is the case, we can efficiently synthesize such exotic peptides containing AcPro at the N-terminus in the presence of EF-P. Howe v er, in the preceding reports, only fMet was evaluated as the N-terminal substrate; Noncanonical, less reacti v e peptidyl donor substrates, such as AcPro, have not been thoroughly examined for EF-P-assisted enhancement of peptide bond formation ( 32 , 33 ). Ther efor e, we first aimed at confirming that EF-P is able to enhance peptide bond formation between the Nterminal AcPro and the second amino acid, thereby alleviating N-terminal drop-off-reinitiation.
We previously reported that EF-P recognizes the specific D-arm structure of tRN A Pro isoacceptors, tRN A Pro1 , tRN A Pro2 and tRN A Pro3 , for acceleration of peptidyl transfer of Pro ( 35 ). Ther efor e, if the D-arm of tRNA ini is substituted with those of tRNA Pro isoacceptors, peptidyl transfer between the N-terminal AcPro and the second amino acid could be further enhanced. Here we aim at de v eloping such an engineered tRNA ini variant by merging the structural features of wild-type tRNA ini and tRNA Pro isoacceptors for more ef ficient incorpora tion of AcPro a t the N-terminus (Figure 1 C). This novel initiator tRNA was r eferr ed to as tRNA iniP . Since we have also reported that IF3, EF-G and RRF are involved in suppression of the N-terminal dropof f-reinitia tion ( 27 ), the concentrations of these translation factors are also optimized for AcPro incorporation. Consequently, the use of tRNA iniP charged with AcPro as well as other exotic nonproteinogenic amino acids, e.g. D -amino, ␤amino and ␥ -amino acids, under the optimized conditions for the protein factors enabled us to efficiently express the full-length peptides including such amino acids at the Nterminus. We also demonstrated macrocyclization of peptides by introducing N-terminal N -chloroacetylated amino acids, which reacted with a sulfhydryl group of downstream Cys to form a thioether bond.

Pr epar ation of tRNA ini variants and flexizymes
Template DNAs for tRNA ini variants and flexizymes (dFx and eFx) were prepared by extension of forward and re v erse extension primer pairs, and PCR using forward and reverse PCR primer pairs (See Supplementary Table S1 for details). Transcription of tRNA ini and flexizymes was conducted at 37 • C for overnight in 250 l and 2 ml, respectively, of the following r eaction mixtur es: 40 mM Tris-HCl (pH 8.0), 22.5 mM MgCl 2 , 1 mM dithiothreitol (DTT), 1 mM spermidine, 0.01% Triton X-100, 3.75 or 5 mM nucleoside triphosphate (NTP) mix, 5 or 0 mM guanosine monophosphate (GMP), 0.04 U / l RNasin RNase inhibitor (Promega) and 0.12 M T7 RN A pol ymerase. 200-lor 2-ml-scale PCR products were added to the above r eaction mixtur e for transcription of tRNA ini and fle xizymes, respecti v ely. The concentrations of NTP mix were 3.75 mM for tRNA ini and 5 mM for flexizymes, and those of GMP were 5 and 0 mM, respecti v ely. The transcribed tRNA ini and flexizymes were treated with RQ1 DNase (Promega) at 37 • C for 30 min and purified on 8% (tRNA ini ) or 12% (flexizymes) polyacrylamide gels containing 6 M urea. The resulting RNAs were eluted from the gel, precipitated by ethanol and dissolved in water.

Pr epar ation of aminoacyl-tRNAs
Aminoacylation of tRNA ini variants was carried out at 0˚C in a following r eaction mixtur e: 50 mM HEPES-KOH (pH 7.5), Bicine-KOH (pH 9.0) or CHES-KOH (pH 10.0), 600 mM MgCl 2 , 20% DMSO, 25 M dFx or eFx, 25 M tRNA ini variants and 5 mM activated amino acids. L-pr oline -3,5-dinitr obenzyl ester (Pr o-DBE), N -acetyl-D -tyrosine-cyanomethyl ester (Ac D Tyr-CME), N -acetyl-D -tryptophane-cyanomethyl ester (Ac D Trp-CME), L -␤-homophenylglycine -3,5-dinitrobenzyl ester ( ␤ Phg-DBE) and 3-aminobenzoic acid-cyanomethyl ester ( 3 Abz-CME) were utilized as the activated amino acids for charging Pro, Ac D Tyr, Ac D Trp, ␤ Phg and 3 Abz, respecti v ely. The reaction was conducted for 2, 3, 3, 22 and 144 h, respecti v ely. Acylations of ␤ Phg and 3 Abz were performed at pH 9.0 and 10.0, respecti v ely, and the other substra tes were acyla ted a t pH 7.5. These activa ted amino acids were synthesized by previously reported methods (36)(37)(38). dFx was used for DBEs and eFx for CMEs. The resulting aminoacyl-tRNAs were precipitated by ethanol, washed twice with 70% ethanol containing 0.1 M sodium acetate (pH 5.2), and dissolved in 1 mM sodium aceta te (pH 5.2). For N -acetyla tion of Pro-tRNA ini , ␤ Phg-tRN A ini and 3 Abz-tRN A ini , 250 pmol aminoacyl-tRN A ini was dissolved in 60 l 0.3 M sodium acetate / 0.5 M acetic anhydride solution (pH 5.2), incubated for 30 min at 25˚C, and then r ecover ed by ethanol precipitation. For N -chloroacetylation of Pro-tRNA ini and 3 Abz-tRNA ini , 500 pmol aminoacyl-tRNA ini was dissolved in 75 l 0.3 M sodium acetate / 40 mM chloroacetic anhydride solution (pH 5.2), incubated for 5 min at 25˚C and r ecover ed Figur e 1. EF-P reco gnizes the D-arm of P-site tRN A f or acceleration of peptide bond f orma tion. ( A ) EF-P-media ted accelera tion of peptide bond formation between two consecuti v e Pro residues at P and A site in transla tion elonga tion ( 30 , 31 ). EF-P recognizes the specific D-arm motif of the P-site peptidylprol yl-tRN A Pro for the acceleration ( 35 ). ( B ) EF-P mediated acceleration of peptide bond formation between P-site fMet-tRNA and A-site amino acid ( 32 , 33 ). This e v ent occurs a t transla tion initia tion. ( C ) De v elopment of a nov el initia tor tRNA tha t can be ef ficientl y reco gnized by EF-P for acceleration of peptide bond formation. The nucleotide sequence of tRNA iniWT is identical to that of E. coli tRNA fMet2 except for the nucleotide modifications due to in vitro transcription of tRNA iniWT . The 5'-end C is replaced by G for tRNA iniG1 . The D-arm motif of tRNA Pro1 can be recognized by EF-P (indicated by red dotted line). By combining the structural features of tRN A iniWT and tRN A Pro1 , the engineered tRN A, tRN A iniP , was de vised. The nucleotides deri v ed from tRNA iniG1 are indicated by blue, those from tRNA Pro1 by red, and the common ones in black. Note that nucleotide modifications are omitted in this figure. by ethanol precipitation. The resulting N -acetyl-or Nchloroacetyl-aminoacyl-tRNAs were washed twice with 70% ethanol containing 0.1 M sodium acetate (pH 5.2) and dissolved in 1 mM sodium acetate (pH 5.2).

T r anslation of model peptides
The model peptide P1 was translated using template DNAs that encode mRNAs mR1 − mR6. The template DNAs wer e pr epar ed by e xtension of forwar d and re v erse e x-tension primer pairs, and PCR using forward and reverse PCR primer pairs (see Supplementary Table S1 for details). Translation was carried out at 37 • C for 30 min in a 2.5 l-scale FIT system of the following composition unless otherwise designated: 50 mM HEPES-KOH (pH 7.6), 100 mM potassium acetate, 12.6 mM magnesium acetate, 2 mM ATP, 2 mM GTP, 1 mM CTP, 1 mM UTP, 20 mM crea tine phospha te , 2 mM spermidine , 1 mM DTT, 1.5 mg / ml Esc heric hia coli total tRNA, 1.2 M E. coli ribosome, 0.6 M methionyl-tRNA formyltr ansfer ase, 2.7 M  Figure S2. Note that the internal control peptides were not added in the experiments shown in Figure 5 , where U-13 C:U-15 N-Lys was utilized for the translation in the presence of EF-P and unlabeled Lys for that in the absence of EF-P.

MALDI-TOF mass spectrometry of model peptides
Translated peptides were desalted with SPE C-tip (Nikkyo Technos) and cocrystallized with ␣-cyano-4hydroxycinnamic acid on a sample plate. For the analysis shown in Figure 5 , two translation solutions deri v ed from EF-P(+) and EF-P( −) experiments were mixed together, and then subjected to C-tip and cocrystallization. MALDI-TOF mass spectrometry (MS) was performed by UltrafleXtreme (Bruker Daltonics) in reflector-positi v e mode. A peptide calibration standard II (Bruker Daltonics) was used for external mass calibration.

Ribosomal incorporation of N -acetyl-L -proline at the Nterminus
To observe N-terminal drop-of f-reinitia tion, AcPro was introduced into a model peptide P1 using an mRNA mR1 (Figure 2 A). For incorporation of AcPro, we first tested three initiator tRN A variants, tRN A iniWT , tRN A iniG1 and tRNA iniG1 / C11 / G24 (Figure 1 C for tRNA iniWT and tRN A iniG1 , 2C for tRN A iniG1 / C11 / G24 ). All of these tR-NAs are deri v ed from E. coli tRNA fMet2 but lack nucleotide modifications due to in vitro transcription. In addition, tRN A iniG1 and tRN A iniG1 / C11 / G24 have G1 and G1 / C11 / G24 mutations, respecti v ely. G1 mutation aimed at improving transcription efficiency and C11 / G24 mutation for efficient recognition by EF-P. Since EF-P recognizes D-arm of tRNA Pro1 and the only difference in the D-arm of tRN A fMet2 and tRN A Pro1 is found at this position (Figure 1 C, A11 / U24 for tRNA fMet2 and C11 / G24 for tRNA Pro1 ) ( 35 ), we assumed that C11 / G24 mutation would enhance its recognition by EF-P. Pro was precharged on these tRNAs using dFx, one of flexizyme variants ( 39 ), and N -acetylated by acetic anhydride to pr epar e AcPro-tRNA.
Translation was conducted in an E. coli reconstituted translation system, r eferr ed to as the fle xib le in vitro translation (FIT) system ( 40 ), containing U-13 C:U-15 N-Lys instead of unlabeled Lys. Thus, the translated peptides were labeled with four U-13 C:U-15 N-Lys (Figure 2 A, B, indicated by green). Consequently, both full-length P1 (P1-FLP) and reinitiated peptide lacking the N-terminal AcPro (P1-RiP) were detected by MALDI-TOF MS (Figure 2 B, a r epr esentati v e r esult for tRNA iniG1 / C11 / G24 in the pr esence of EF-P). The e xpression le v els of the translated P1-FLP and P1-RiP wer e estimated by their r elati v e peak intensities to those of 0.5 M synthetic internal control peptides, contr ol-P1-FLP and contr ol-P1-RiP, bearing unlabeled Lys. We assumed that the translated and control P1 peptides have equal ionization efficiencies due to the identical amino acid sequences except for the isotope labeling. Consequently, the le v els of P1-FLP using tRNA iniWT , tRN A iniG1 and tRN A iniG1 / C11 / G24 were 0.067, 0.061 and 0.87 M, respecti v el y, in the presence of EF-P; w hile 0.045, 0.038 and 0.23 M, respecti v ely, in the absence of EF-P (Figure 2 D). The percentages of P1-FLP [P1-FLP%: P1-FLP / (P1-FLP + P1-RiP) × 100] were 24%, 23% and 43%, respecti v ely, in the presence of EF-P and 17%, 17% and 34%, respecti v ely, in the absence of EF-P (Figure 2 D). Both P1-FLP le v el and P1-FLP% were enhanced by EF-P for all of these tRNAs, indicating that their D-arms could be recognized by EF-P. In addition, tRNA iniG1 / C11 / G24 exhibited 14-fold higher expression level of P1-FLP and 20% higher P1-FLP% compared to the use of tRNA iniG1 in the presence of EF-P (0.87 M versus 0.061 M and 43% versus 23%), showing that the introduction of the C11 / G24 mutation contributed to enhancement of the P1-FLP le v el and P1-FLP%. Since the enhancement effects were observ ed e v en in the absence of EF-P for tRNA iniG1 / C11 / G24 , it is likely that its conformational change induced by the C11 / G24 mutation is preferable for AcPro incorporation. On the other hand, the G1 mutation did not affect the function of initiator tR-NAs. Ther efor e, we decided to introduce G1 and C11 / G24 mutations to all initiator tRNAs hereafter.

Screening of tRNA ini variants for efficient expression of the full-length peptide
The above result motivated us to further optimize the local structures of tRNA iniG1 / C11 / G24 to improve the efficiency of AcPro incorporation into full-length peptides. P1-FLP le v el and P1-FLP% in the presence of EF-P were used as benchmarks for evaluation of tRNA ini variants. We screened such a tRNA ini variant that shows the highest P1-FLP le v el and P1-FLP% so tha t we can ef ficiently and cleanly express full-length P1-FLP bearing AcPro at the N-terminus using the tRNA ini variant. Here we introduced fiv e anticodon stem variations (An1 −5), fiv e acceptor stem variations (Ac1 −5), four T-stem variations (T1 −4), four variable loop variations (V1 −4), and their combinations, where the sequence of tRNA Pro1 was partially implanted into tRNA iniG1 / C11 / G24 (Figure 3 , Supplementary Figure S1, bases deri v ed from tRNA iniG1 are indicated in b lue, those from tRNA Pro1 in red, and common bases in black). Note that tRNA iniG1 / C11 / G24 has An1, Ac1, T1 and V1 and is referred to as tRNA An1Ac1T1V1 hereafter (Figure 2 C); other mutants are also named similarly after their local structures: tRN A AnXAcXTXVX , w here X indicates the numbering of local structures shown in Figure 3 A. First, we evaluated anticodon stem variants (Figure 3   The anticodon stem, acceptor stem, T-stem and variable loop of this tRNA ar e r eferr ed to as An1, Ac1, T1 and V1, respecti v el y. This tRN A is also called tRNA An1Ac1T1V1 . ( D ) Quantification of the expression levels of P1-FLP and P1-RiP. Translation of these peptides was conducted using tRNA iniWT , tRN A iniG1 and tRN A An1Ac1T1V1 in the presence and a bsence of EF-P. Numbers a bove the bars indicate P1-FLP%. n = 3. Error bars , S .D. S1A, tRNA An2 −5Ac1T1V1 ). Among them, tRNA An4Ac1T1V1 showed the highest P1-FLP le v el and P1-FLP% in the presence of EF-P (Figure 3 B, 0.98 M and 58%). Thus, we chose An4 as the best anticodon stem structure. Second, we evaluated acceptor stem variants, where anticodon stem was fixed to An4 (Figure 3 C, Supplementary Figure 1B, tRN A An4Ac2 −5T1V1 ). Consequentl y, none of these variants showed higher P1-FLP le v el nor P1-FLP% than those of tRNA An4Ac1T1V1 (Figure 3 C, 0 −0.24 M and 0 −14%, respecti v ely, in the presence of EF-P). Thus, we decided to keep using Ac1 as the best acceptor stem among Ac1 − 5. Third, combinations of T-stem variations and variable loop variations were evaluated using tRN A An4Ac1T2 −4V2 −4 , w here anticodon and acceptor stems were fixed to An4 and Ac1, respecti v ely (Figure 3 D, Supplementary Figure S1C). For the T-stem variants, T3 showed generally higher P1-FLP le v el and P1-FLP% than T1, T2 and T4 (Figure 3 D). For the variable loop variants, V3 showed higher P1-FLP le v el and P1-FLP% than V1, V2 and V4 (Figure 3 D). Thus, the combination of T3 and V3 resulted in the highest P1-FLP le v el as well as P1-FLP% (Figure 3 D, tRNA An4Ac1T3V3 , 1.12 M and 55%, respecti v ely, in the presence of EF-P). Since tRNA An4Ac1T3V3 showed the highest P1-FLP le v el among all tRNAs evaluated in this study, we decided to use this tRNA for further validations. tRNA An4Ac1T3V3 is referred to as tRNA iniP hereafter (Figure 1 C, Supplementary  Figure S1C).

Optimization of translation conditions for efficient expression of the full-length peptide
We r ecently r e v ealed tha t the N-terminal drop-of fr einitiation is suppr essed by IF3, EF-G and RRF ( 27 ). Ther efor e, her e we optimized the concentrations of IF3, EF-G and RRF to enhance the P1-FLP le v el and P1-FLP% (Figure 4 A-C). In addition, concentrations of EF-P and AcPro-tRNA iniP were also titrated for the P1-FLP synthesis (Figure 4 D, E). In titration of IF3, P1-FLP le v el plateaued at over 5 M IF3 and P1-FLP% surpassed 90% at over 15 M (Figure 4 A). Thus, we decided to use 15 M IF3 for the rest of the e xperiments. In EF-G titration, P1-FLP le v el peaked at 1 M EF-G with 98% P1-FLP% and gradually decreased at the higher EF-G concentrations above 1 M (Figure 4 B), indicating the possibility that too high concentration of EF-G induces frequent mistranslocation and drop-off of AcPro-tRNA iniP from the P site. For RRF and AcPro-tRNA iniP concentrations, P1-FLP le v el pla teaued a t over 1 M RRF and 80 M AcPro-tRNA iniP , respecti v ely (Figure 4 C,E). For EF-P concentration, P1-FLP le v el peaked at 5 −10 M EF-P and gradually decreased at higher EF-P concentrations (Figure 4 D). This is likely because such high concentrations of EF-P remain to occupy the ribosomal E site e v en after the peptidyl transfer completes and thus inhibit the translocation of deacyl-tRNA from P site to E site. We previously observed the same tendencies for elongation of inefficient substrates such as Pro, D -amino-, ␤-amino-, and ␥ -amino acids ( 15 , 22 , 23 , 35 , 41 ). Comparing the P1-FLP le v el at 10 M EF-P with that at 0 M EF-P, 2.8-fold improvement was observed (Figure 4   Since we recently found that AUG is not necessarily the best initiator codon for incorporation of AcPro but other codons, e.g. AAG, possibly show higher P1-FLP% ( 27 ), we tested nine codon-anticodon combinations using tRNA iniP for AcPro incorporation into P1 (Supplementary Figure Table  S1). Note that the concentration of AcPro-tRNA iniP was increased to 160 M in this analysis. As a result, A UU / GA U, A UC / GA U, AAG / CUU and GU A / U AC showed significantly higher P1-FLP le v el than the canonical A UG / CA U (Supplementary Figure S2A Thus, we concluded that AAG / CUU and GU A / U AC ar e pr eferable codon-anticodon combinations for P1-FLP synthesis. We also evaluated the effect of Shine-Dalgarno (SD) sequence and the spacer between SD and initiation codon. In addition to mR1, fiv e SD + spacer sequences were evaluated, where AUG and AAG were introduced as initiation codons (Supplementary Figure S2B, mR2 −mR6). As a result, we observed a wide range of P1-FLP levels depending on the type of SD + spacer, where mR5 exhibited the highest P1-FLP le v el for both AUG and AAG codons [Supplementary Figure S2C

Ribosomal incorporation of D -amino, ␤-amino and ␥-amino acids at the N-terminus by means of tRNA iniP and EF-P
Next, we aimed at a ppl ying tRN A iniP for incorporation of D -amino, ␤-amino and ␥ -amino acids using N -acetyl-D -tryptophan (Ac D Trp), N -acetyl-D -tyrosine (Ac D Tyr), Nacetyl-L -␤-homophenylglycine (Ac ␤ Phg) and N -acetyl-3aminobenzoic acid (Ac 3 Abz) as their r epr esentati v es (Figure 5 A). These amino acids were precharged on tRNA iniP bearing CUU anticodon, r eferr ed to as tRNA iniP CUU and introduced at the N-terminus of P1 peptide using mR5-AAG to yield P1-FLP-Ac D Trp, P1-FLP-Ac D Tyr, P1-FLP-Ac ␤ Phg and P1-FLP-Ac 3 Abz. Translation reaction was carried out at 37˚C for 30  The N -chloroacetyl group spontaneously reacted with the sulfhydryl group of a downstream Cys to form an irreducible thioether bond, resulting in macrocyclization of peptide ( 42 ). MALDI-TOF MS of the translation products showed that the desired macrocyclic peptides were cleanly expressed without truncation ( Figure 6 D-G, P7-FLP-ClAcPro and P7-FLP-ClAc 3 Abz).

DISCUSSION
Here, we have demonstra ted tha t EF-P is able to enhance incorporation of di v erse N-terminal substrates. We previously reported that EF-P recognizes the specific D-arm motif of tRNA Pro isoacceptors comprised of a 9-nt D-loop closed with a 4-bp D-stem, where high GC content of Dstem is r equir ed for efficient recognition ( 35 ). The D-arm of tRNA fMet2 and tRNA Pro1 are almost identical except for one base pair at position 11 and 24 of the D-stem, where an A / U base pair is found in tRNA fMet2 and a C / G pair in tRNA Pro1 . Ther efor e, it is r easonable that substitution of the A / U pair of tRN A iniG1 , w hose D-arm is deri v ed from tRNA fMet2 , with a C / G pair resulted in further improvement of the P1-FLP le v el and P1-FLP% in the presence of EF-P. The pyrimidine 11 / purine 24 pair is unique and widely conserved in prokaryotic initiator tRNAs ( Figure  1 C) ( 43 , 44 ); Howe v er, it was reported that a mutant initiator tRNA bearing C / G pair at this position was quite acti v e in protein synthesis in vi vo ( 45 ). Similar l y, the C / G m utation in our engineered initiator tRNAs was tolerated for AcPro incorporation.
By using the engineered initiator tRN A, tRN A iniP , under the optimized translation conditions, we succeeded in efficient incorporation of AcPro at the N-terminus with 1000fold improvement of expression level compared to that of our conventional conditions using tRNA iniWT . Moreover, N-terminal drop-of f-reinitia tion e v ent was completely suppressed under these conditions. The N-terminal Pro residue is an attracti v e building b lock of bioacti v e foldamer peptides owing to its constrained cyclic structure that contributes to stabilization of turn and helical conformations ( 1 , 2 ). D -Amino, ␤-amino and ␥ -amino acids are also useful building blocks of bioactive peptides that can be introduced by our methodology. We succeeded in introducing Ac D Trp, Ac D Tyr, Ac ␤ Phg and Ac 3 Abz into model peptides as their r epr esentati v es. We can expect their unique and strong folding propensities, such as turn / helix inducing abilities (3)(4)(5)(6)(7)(8)(9)(10)(11)(12). Macrocyclization is also a powerful approach to construct constrained geometries of peptides. We showed that tRNA iniP is applicable to introduction of N -chloroacetyla ted substra tes, ClAcPro and ClAc 3 Abz, at the N-terminus for macrocyclization of peptides ( Figure 6 ). We can expect high binding affinity and inhibitory activity against target molecules, improved membrane permeability, and proteolytic stability for these foldamer peptides (13)(14)(15)(16)(17). Peptides consisting of only L -␣-amino acids are ra pidl y degraded by peptidases in serum or in cells, which is often a critical disadvantage of peptide drugs. Howe v er, by introducing noncanonical amino acids at the N-terminus, their proteolytic stability against exopeptidases drastically improves ( 46 , 47 ). In addition, their N -acetylation also increases proteolytic stability ( 48 ). The advantage of  ribosomally synthesizing such exotic peptides is that, by translating mRNAs with random sequences, we can easily pr epar e random peptide libraries, which can be applied to display-based screening methodologies, such as the RaPID (Random nonstandard Peptides Integrated Discovery) system ( 40 ). RaPID screening of novel bioactive peptides bearing exotic amino acids at the N-terminus would be performed in our future studies.

DA T A A V AILABILITY
The data underlying this article are available in the article and in its online supplementary material.